Manu Shukla

Posted on Jul 21 • Originally published at ecorpit.com

HRMS build vs buy 2026: the 50% wage rule and what each option really costs

#hrms

HRMS build vs buy 2026: the 50% wage rule and what each option really costs

Summary. India's four Labour Codes came into force on 21 November 2025, rationalising 29 central labour laws in one step, and the Code on Wages, 2019 redefined what counts as "wages" for statutory purposes. Excluded allowances can no longer exceed 50% of total remuneration, which pushes basic pay up for every salary structure built the old way and raises the base for provident fund at 12%, ESI at 3.25% employer plus 0.75% employee, gratuity and bonus. Legal analysis puts the resulting statutory cost increase at 5% to 15% depending on existing structure. Meanwhile Zoho Payroll's India plans start free for up to 10 employees and reach ₹4,000 per organisation per month plus ₹80 per additional employee on annual billing, exclusive of GST. That gap, between a ₹1.92 lakh annual subscription and a bespoke system, is the decision most HR and engineering leaders are now re-opening.

The honest answer is that most companies should buy, and the ones that should build usually know why before they ask. What has changed in 2026 is that the buy option got harder to configure and the build option got a permanent maintenance line item.

What actually changed on 21 November 2025

The Ministry of Labour and Employment made all four Codes effective on the same day: the Code on Wages, 2019, the Industrial Relations Code, 2020, the Code on Social Security, 2020 and the Occupational Safety, Health and Working Conditions Code, 2020. The Press Information Bureau release describes it as rationalising 29 existing central labour laws.

Several of those changes are direct software requirements, not policy statements. An HRMS that cannot produce them is non-compliant regardless of how good its dashboards look.

Change under the Codes	What the system must now do	Where it bites
Mandatory appointment letters for all workers	Generate, version and store a letter per worker stating designation, wages and social security entitlements	Onboarding module, document store, audit trail
Gig and platform work defined for the first time	Track aggregator contributions of 1% to 2% of annual turnover, capped at 5% of amounts paid to platform workers	Contractor and vendor payments, not payroll
Fixed-term employees get gratuity after one year	Accrue gratuity from year one for FTEs, not year five	Gratuity engine, separation workflow
ESIC extended pan-India	Voluntary below 10 employees, mandatory where even one worker is in a hazardous process	Eligibility rules per establishment, not per company
Single registration, single licence, single return	Consolidate filings that were previously separate per statute	Compliance calendar and filing integrations
Salary release by the 7th for IT and ITES workers	Hard payroll cut-off with no slippage	Pay run scheduling and approvals
Free annual health check-up for workers over 40	Track age-based benefit eligibility and completion	Benefits and vendor scheduling

The Codes also formalise a broader shift the PIB release quantifies: social security coverage rose from about 19% of the workforce in 2015 to more than 64% in 2025. More covered workers means more statutory calculation, per head, per month.

The wage definition is the expensive part

Section 2(y) of the Code on Wages defines wages as basic pay, dearness allowance and retaining allowance. It then caps the excluded components, house rent allowance, conveyance and other special allowances, at 50% of total remuneration. Anything above that cap is pulled back into wages for calculation.

Indian salary structures have been engineered against exactly this for two decades. A basic component of 20% to 30% of CTC, with the rest in allowances, was the standard way to hold down the base on which PF, ESI, gratuity and bonus are computed. That structure no longer works.

Run the arithmetic on a single employee. Take ₹10,00,000 CTC with basic at 30%, so ₹3,00,000. Employer PF at 12% of basic is ₹36,000. Restructure to meet the definition and basic becomes ₹5,00,000, so employer PF becomes ₹60,000, before EDLI at 0.50% and administrative charges at another 0.50%. Gratuity liability, calculated on last drawn basic, rises in the same proportion. Legal commentary on the rule puts the aggregate statutory increase at 5% to 15% for most employers, concentrated in IT services, business process outsourcing, retail and hospitality, where low-basic structures were most common.

For the software, the consequence is specific. Every salary template, every CTC calculator, every offer-letter generator, every arrears computation and every historic payslip regeneration path has to be rebuilt against a new definition of wages. This is not a configuration change. It is a change to the primitive that the rest of payroll is derived from.

The statutory rates themselves did not move, which is worth stating plainly because a lot of 2026 commentary implies otherwise:

Contribution	Rate	Applicability threshold
EPF, employee	12% of wages	Wage ceiling of ₹15,000 per month for mandatory coverage
EPF, employer	12%, split 3.67% to EPF and 8.33% to EPS	Plus 0.50% EDLI and 0.50% administrative charges
ESI, employee	0.75% of gross wages	Gross up to ₹21,000 per month
ESI, employer	3.25% of gross wages	Establishments with 10 or more employees in most states

The rates are stable. The base they apply to is what changed, and the base is computed in software.

What buying actually costs in 2026

Published Indian pricing gives a firm floor for the buy case. Zoho Payroll's India pricing page, checked on 21 July 2026, lists a free tier at ₹0 for up to 10 employees that still includes income tax, EPF, ESI, state-wise professional tax and labour welfare fund compliance. Paid tiers on annual billing run ₹1,000 per organisation per month for 25 employees, ₹3,000 for 50, and ₹4,000 for 50 on the Premium plan, with additional employees at ₹40, ₹60 and ₹80 respectively. All prices exclude GST.

That produces concrete annual numbers. A 200-person company on Premium annual billing pays ₹4,000 plus 150 additional employees at ₹80, so ₹16,000 per month, or ₹1.92 lakh per year before GST. Form 16 generation with digital signature, Form 24Q and TDS challan recording sit in the Standard tier and above.

One detail in Zoho's own FAQ matters for group structures: a separate payroll licence is required per legal entity, though a single licence covers multiple branches of one entity. Companies running three or four Indian entities should price accordingly.

Other Indian platforms publish comparable shapes. Reported 2026 pricing puts Zoho People's HR tiers between ₹50 and ₹230 per employee per month, Keka's plans at ₹9,999 to ₹15,999 per month with additional heads at ₹90 to ₹150, and greytHR from roughly ₹3,495 per month at 50 employees. Darwinbox quotes on scope rather than publishing rates.

Option	Indicative annual cost, 200 employees	What you are actually buying
SaaS HRMS, mid tier	Roughly ₹1.9 lakh to ₹5 lakh, ex-GST	Statutory updates shipped by the vendor, fixed data model
SaaS plus integration work	Subscription plus one-time integration engineering	Same, with your ERP, biometrics and finance systems wired in
Custom build	Quoted per scope, plus a permanent compliance maintenance line	A data model that matches how you actually employ people
Hybrid	Bought payroll core, built everything around it	Vendor absorbs statutory churn, you own the differentiated workflow

We are not going to publish a single build price, because a number that does not name the scope is marketing rather than information. What we can say is which variables move it: the number of legal entities, the number of states with distinct professional tax and labour welfare fund rules, whether attendance comes from biometric hardware or an app, whether payroll integrates with an existing ERP, and how much historic data has to migrate with recomputation.

When building is the right call

Most companies under 500 employees should buy. The statutory surface is wide, it changes without warning, and a vendor amortises that maintenance across thousands of customers. Building a payroll engine to save a subscription of a few lakh a year is a poor trade.

Four situations genuinely justify a build, and they share a common shape: the constraint is structural, not financial.

Workforce models the packaged tools do not represent. Companies running large contractor networks, per-shift or per-piece pay, or blended gig and payroll workforces routinely find that off-the-shelf HRMS data models cannot represent a worker who is simultaneously on two engagement types. The Codes making gig and platform work a defined category, with aggregator contributions of 1% to 2% of turnover, has made this more common rather than less.

HR as a product surface. If your employee app is customer-facing or field-facing, a hospital's rostering app, a logistics operator's driver app, a retail chain's shift-swap app, then HR data is part of your product and it needs to live where your product lives. This is a mobile engineering problem more than an HR problem, and it is the case we see most often. Our guide to enterprise mobile app development covers the architecture side of this.

Data residency and access constraints. Some regulated employers cannot put employee master data in a multi-tenant platform. That is a build or a private deployment, not a preference.

Integration cost exceeding build cost. When a company already runs a heavily customised ERP and the integration work to make a SaaS HRMS fit approaches the cost of building the modules natively, the calculation flips. This is rarer than vendors of custom software suggest and more common than SaaS vendors admit.

If none of those apply, buy the payroll core and spend the engineering budget on the layer above it, which is where the differentiation actually is.

DPDP obligations on employee data

Employee records are among the most sensitive personal data any Indian company holds, and the Digital Personal Data Protection Act, 2023 with the DPDP Rules notified in November 2025 now governs them directly. Full compliance is required by mid-May 2027 after an 18-month implementation phase, and penalties run to ₹250 crore.

The useful part for an HRMS decision is what the Act does not require. Processing employee data to fulfil employment obligations and statutory compliance, PF, ESI and TDS among them, falls under legitimate uses and does not need separate consent. Processing outside that, using employee data for marketing or sharing it with third parties, does.

Three design requirements follow, and they apply whether you build or buy:

Biometric attendance needs explicit handling. Fingerprint and retina data used for attendance is sensitive personal data. It needs informed consent before collection, encryption at rest, deletion when the purpose ends, which for most employers means at exit, and no third-party sharing without fresh consent. A biometric attendance system with no deletion path is a liability, and a surprising number of deployed systems have none.

Notice and withdrawal have to be real. Notices must be itemised and separate from general terms, in clear language, available in English or a Schedule VIII language, with a working mechanism to withdraw consent, exercise rights and raise grievances. Withdrawal must be as easy as consent was.

Erasure and breach paths must exist in the schema. Personal data has to be erasable on consent withdrawal or when the retention purpose ends, and breaches must be notified to the Data Protection Board and to each affected person. Retrofitting erasure into a payroll schema with a decade of denormalised history is the expensive version of this. We wrote up that specific problem in DPDP legacy data remediation and consent backfill, and the broader engineering view in the DPDP engineering playbook for Indian startups.

A realistic delivery timeline

For a custom or hybrid HR platform, the sequence below reflects how these programmes actually run, not how they are pitched. Durations depend on entity count and integration surface, so treat the ordering as the fixed part.

Statutory model first. Define wages under Section 2(y), the contribution rules, and the state-wise professional tax and labour welfare fund matrix before any interface work. Getting this wrong late is the most expensive failure mode in HR tech.
Employee master and org structure. Legal entity, location, establishment, cost centre and reporting line, with effective-dated history from the start. Retrofitting effective dating is close to a rewrite.
Attendance and leave. Source of truth decisions here determine everything downstream. Biometric integration brings the DPDP obligations above with it.
Payroll engine and pay runs. Salary templates, arrears, off-cycle runs, salary hold, loans and advances, statutory bonus.
Filings and documents. Form 16 with digital signature, Form 24Q, TDS challans, PF and ESI returns, the compliance calendar.
Self-service and mobile. Payslips, declarations, reimbursements, approvals. This is the part employees judge the system by, and it is deliberately last.
Parallel run. Two to three cycles running old and new side by side, reconciling to the rupee before cutover. Nobody regrets a parallel run.

The step teams skip is the parallel run, and it is the step that catches the wage-definition errors before an employee's take-home pay is wrong.

What this does to take-home pay, and why HR needs to plan for it

There is a communication problem sitting inside the technical one. When basic pay rises to meet the definition, the employee's own PF contribution at 12% of basic rises with it. Total compensation is unchanged and the retirement corpus grows faster, but monthly in-hand pay falls for anyone previously on a 30% basic structure.

Employees will notice the smaller number before they appreciate the larger corpus. An HRMS rollout that changes salary structure without a payslip breakdown showing the before and after, and without a modelled projection of the PF gain, converts a compliance exercise into a retention problem. Build the comparison view. It costs a sprint and it prevents a quarter of noise.

How eCorpIT can help

eCorpIT is a Gurugram technology consultancy founded in 2021, working at CMMI Level 5 and MSME certified, with partnerships including AWS, Microsoft and Google. We build HR and workforce platforms for Indian companies whose employment model does not fit a packaged product, and we are equally willing to tell you to buy Zoho or Keka and spend the budget elsewhere, because that is the right answer more often than not. Where we do build, our senior-led teams design applications aligned with DPDP Act requirements from the schema up, including erasure paths and consent records, and we run the parallel cycles before cutover rather than after. If you are weighing a build against a subscription, talk to us about a scoped HR platform assessment and we will size both honestly. Our mobile app development RFP template is a reasonable starting point if you are going out to tender.

FAQ

What changed for Indian payroll on 21 November 2025?

The Government made all four Labour Codes effective that day, rationalising 29 central labour laws. The Code on Wages redefined wages as basic pay plus dearness allowance and retaining allowance, and capped excluded allowances at 50% of total remuneration, changing the base for provident fund, ESI, gratuity and bonus.

Does the 50% wage rule increase employer cost?

Yes, for most employers. Raising basic pay raises the base for employer provident fund at 12%, ESI at 3.25%, gratuity and bonus. Legal commentary estimates aggregate statutory cost increases of 5% to 15%, concentrated in IT services, business process outsourcing, retail and hospitality where low-basic structures were common.

Should we build our own HRMS or buy one?

Most companies under 500 employees should buy. Building is justified when your workforce model cannot be represented in packaged tools, when HR data is part of a customer-facing product, when data residency rules prevent multi-tenant hosting, or when integration cost with an existing ERP approaches the cost of building natively.

How much does HRMS software cost in India in 2026?

Zoho Payroll's India page lists a free tier for up to 10 employees and annual-billing plans from ₹1,000 to ₹4,000 per organisation per month, plus ₹40 to ₹80 per additional employee, excluding GST. A 200-person company on the Premium plan pays about ₹1.92 lakh per year.

Do we need one payroll licence per legal entity?

Under Zoho Payroll's published terms, yes. A separate licence is required for each legal entity, while multiple branches of a single entity can run on one licence. Group structures with three or four Indian entities should price the subscription accordingly before comparing against a build.

What does DPDP require for employee data?

Processing for employment obligations and statutory compliance such as provident fund, ESI and TDS falls under legitimate uses and needs no separate consent. Anything beyond that does. Biometric attendance data needs explicit consent, encryption, and deletion once its purpose ends, typically at employee exit.

When is DPDP compliance actually due?

The DPDP Rules were notified in November 2025 with an 18-month implementation phase, putting full compliance around mid-May 2027. Penalties under the Act reach ₹250 crore. Erasure and consent-withdrawal paths are the parts that take longest to retrofit into an existing payroll schema.

What is the most common HR tech implementation mistake?

Skipping the parallel run. Running the old and new systems side by side for two or three cycles and reconciling to the rupee is what catches wage-definition and arrears errors before they reach an employee's bank account. It is also the step most often cut when a programme runs late.